Evaluating LIS Instructional Videos: A Comparative Study of AI and Human Assessments

The integration of artificial intelligence (AI) into educational content evaluation offers a promising avenue for enhancing assessment quality and scalability. This study compares AI-generated evaluations, using ChatGPT, with human assessments for 10 Library and Information Science (LIS) instructional videos from the National Institute of Open Schooling (NIOS). Key evaluation features include video resolution, content relevance, engagement, and clarity. Results show that while AI aligned well with human evaluators for objective features such as resolution and duration, discrepancies were significant in subjective areas like engagement and clarity, with error rates exceeding 75% for some videos. Correlation patterns revealed that content complexity and design flaws influenced evaluation alignment. The study underscores the need for hybrid frameworks combining AI’s efficiency with human expertise to improve instructional video evaluations. Recommendations include refining AI for qualitative assessments and optimising video content design, contributing to more reliable evaluation methods in educational contexts.